chatgpt-pdf.r

pdf_full <- pdftools::pdf_text("pdf_full.pdf")


# 33-36, 38-39, 41-46, 48-49, 52-54, 56-61, 63-72, 74-77, 79-83, 85-91, 93-96, 98-100, 102-105

paginas <- c(
  33:36, 38:39, 41:46, 48:49,
  52:54, 56:61, 63:72, 74:77,
  79:83, 85:91, 93:96, 98:100,
  102:105
)

pegar_infos_pagina <- function(txt_pag, pag) {
  usethis::ui_info("Pegando pagina {pag}...")

  f <- glue::glue("saidas_paginas/{pag}.json")

  if (!fs::file_exists(f)) {
    result <- openai::create_chat_completion(
      model = "gpt-4-1106-preview",
      messages = list(
        list(
          role = "system",
          content = readr::read_file("prompt.md")
        ),
        list(
          role = "user",
          content = txt_pag
        )
      ),
      temperature = 0
    )
    txt_resp <- result$choices[["message.content"]] |>
      stringr::str_remove_all("```json\n|\n```")

    readr::write_file(txt_resp, f)
  } else {
    txt_resp <- readr::read_file(f)
  }

  txt_resp |>
    jsonlite::fromJSON() |>
    tibble::as_tibble()

}

safe <- purrr::possibly(pegar_infos_pagina, tibble::tibble(erro = "erro"))

da_pessoas <- pdf_full[paginas] |>
  purrr::set_names(paginas) |>
  purrr::imap(safe) |>
  purrr::list_rbind(names_to = "pagina")

da_pessoas |>
  dplyr::count(pagina, name = "aff") |>
  dplyr::filter(aff != 16) |>
  print(n = 100)

da_pessoas |>
  writexl::write_xlsx("contatos_pessoas_pdf.xlsx")